Data prediction method and device of radio altimeter, equipment and medium

Abstract

The invention relates to a data prediction method and device of a radio altimeter, equipment and a medium. The method comprises the following steps: acquiring an original height parameter and motion data of a prediction target, and real-time meteorological data of an environment where the prediction target is located; the original height parameters, the motion data and the real-time meteorological data are all provided with time labels; based on the original height parameter and the motion data, quantizing the change condition of the height of the prediction target to obtain a height feature vector; the height feature vector at least comprises a smooth original height parameter with a time label; based on a preset topographic feature library and the height feature vector, determining a topographic type of the prediction target, and obtaining a topographic type label of the prediction target; and based on the terrain type label and the real-time meteorological data, correcting and smoothing the original height parameter to obtain an accurate height prediction parameter. By adopting the method, the reliability and the accuracy of height measurement data of the aircraft in the low-altitude complex terrain flight stage can be improved.

Description

Technical Field

[0001] This invention belongs to the field of flight altitude prediction, and in particular relates to a data prediction method, apparatus, device and medium for a radio altimeter. Background Technology

[0002] With the development of flight altitude prediction technology, radio altitude measurement technology based on multi-source data fusion has emerged. This technology integrates information from multiple sensors such as inertial navigation systems and atmospheric data computers to comprehensively optimize altitude measurement values, thus forming the current mainstream "filter-barometric correction" altitude measurement method.

[0003] In existing altitude measurement methods, the process first uses a Kalman filter algorithm to preprocess the raw altitude parameters output by the radio altimeter, then uses a state-space model to estimate and suppress measurement noise, resulting in a smoothed altitude sequence. Simultaneously, the system acquires static pressure and temperature data from an atmospheric data computer, calculates barometric altitude using a standard atmospheric model, and uses this as a reference. During processing, a sliding time window is established to perform trend analysis on the smoothed radio altitude sequence, then weighted data is fused with barometric altitude data, and finally, an optimized altitude value is output using an adaptive smoothing algorithm.

[0004] However, while current altitude measurement methods can effectively meet the altitude measurement needs in conventional flight environments, their core deficiency lies in their inability to identify and compensate for radio wave multipath propagation errors caused by complex terrain features. When an aircraft flies over complex terrains such as mountains, urban complexes, or undulating hills, the multipath reflections of radio signals on the ground surface with different slopes and reflection characteristics will cause systematic deviations in altitude measurement values. This inherent error pattern determined by terrain features cannot be eliminated by simple data filtering or barometric reference, severely reducing the reliability and accuracy of altitude measurement data during low-altitude flight in complex terrain. Summary of the Invention

[0005] Therefore, it is necessary to provide a data prediction method, apparatus, equipment, and medium for radio altimeters that can improve the reliability and accuracy of altitude measurement data for aircraft during low-altitude complex terrain flight, in order to address the aforementioned technical problems.

[0006] In a first aspect, this application provides a data prediction method for radio altimeters, comprising:

[0007] Acquire the raw altitude parameters and motion data of the target, as well as the real-time meteorological data of the target's environment; the raw altitude parameters, motion data, and real-time meteorological data are all time-stamped; the motion data includes at least vertical velocity; the real-time meteorological data includes at least air pressure and temperature parameters;

[0008] Based on the original height parameters and motion data, the change in the target's height is quantified and predicted to obtain a height feature vector; the height feature vector includes at least the smoothed original height parameters with time labels.

[0009] Based on a pre-defined terrain feature library and height feature vector, the terrain type of the predicted target is determined, and the terrain type label of the predicted target is obtained.

[0010] Based on terrain type labels and real-time meteorological data, the original height parameters are corrected and smoothed to obtain accurate height prediction parameters.

[0011] Furthermore, based on the original height parameters and motion data, the change in the target's height is quantified to obtain a height feature vector, including:

[0012] Based on motion data, the original height parameters are filtered to obtain smoothed original height parameters with time labels.

[0013] Based on the preset sampling time label parameters, the smoothed original height parameters with time labels are combined to obtain the sampling height parameter sequence;

[0014] Based on the sampled height parameter sequence, the mean of each smoothed original height parameter in the sampled height parameter sequence is calculated to obtain the smoothed height mean.

[0015] Based on the sampled height parameter sequence and the smoothed height mean, the change characteristics of the target's height are quantified and predicted to obtain height change characteristic parameters; the height change characteristic parameters include height variance, height change slope, height difference mean, and height difference variance;

[0016] The sampled height parameter sequence is transformed in the frequency domain to obtain frequency domain height data, and the frequency domain features of the frequency domain height data are extracted to obtain frequency domain feature variation parameters; the frequency domain feature variation parameters include spectral centroid and spectral entropy;

[0017] The height feature vector is obtained based on the height feature variation parameters, frequency domain feature variation parameters, and smoothed original height parameters.

[0018] Furthermore, based on a pre-defined terrain feature library and height feature vector, the terrain type of the predicted target is determined, resulting in a terrain type label for the predicted target, including:

[0019] Based on the terrain feature library, the height feature vector is input into the trained terrain classification model to obtain the terrain type and corresponding confidence of the predicted target, and the terrain type label is obtained based on the terrain type and confidence.

[0020] The trained terrain classification model was obtained through the following method:

[0021] Obtain the terrain feature library; the terrain feature library includes the height feature vectors of historical predicted targets and the corresponding terrain type labels;

[0022] The terrain feature database is divided into a model training set and a model validation set according to a preset ratio;

[0023] Based on the model training set and the preset kernel function, an initial terrain classification model is constructed with height feature vector as input and terrain category label as prediction target;

[0024] The accuracy of the initial terrain classification model is verified using a model validation set. When the accuracy of the initial terrain classification model reaches a preset threshold, the terrain classification model is obtained.

[0025] Furthermore, based on terrain type labels and real-time meteorological data, the original height parameters are corrected and smoothed to obtain accurate height prediction parameters, including:

[0026] Based on preset terrain height correction rules and terrain type labels, terrain compensation parameters are obtained; the terrain compensation parameters are labeled with time.

[0027] Based on real-time meteorological data, the impact of real-time meteorological data on smoothing the original height parameter is quantified to obtain meteorological compensation parameters; the meteorological compensation parameters are labeled with time.

[0028] Based on terrain type labels, real-time meteorological data, terrain compensation parameters, and meteorological compensation parameters, the original height parameters are adjusted to obtain accurate height prediction parameters.

[0029] Furthermore, based on real-time meteorological data, the impact of real-time meteorological data on smoothing the original height parameter is quantified to obtain meteorological compensation parameters, including:

[0030] Based on the air pressure and temperature parameters, the refractive index parameters of the environment in which the target is located are calculated.

[0031] The refractive index parameter is gradient calculated to obtain the vertical refractive index gradient; and based on the vertical refractive index gradient, the equivalent Earth radius factor is calculated.

[0032] Meteorological compensation parameters are calculated based on the equivalent Earth radius factor, the preset atmospheric proportion factor, and the smoothed original height parameters.

[0033] Furthermore, based on terrain type labels, real-time meteorological data, terrain compensation parameters, and meteorological compensation parameters, the original height parameters are adjusted to obtain accurate height prediction parameters, including:

[0034] Based on the terrain type label, the confidence score is normalized to obtain the terrain influence weight;

[0035] Based on real-time meteorological data, the fluctuation of real-time meteorological data is quantified to obtain the standard deviation of meteorological data; the standard deviation of meteorological data includes the standard deviation of air pressure and the standard deviation of temperature.

[0036] The meteorological impact weights are calculated based on the standard deviation of meteorological data and real-time meteorological data.

[0037] Based on the meteorological influence weight, topographic influence weight, topographic compensation parameter, and meteorological compensation parameter, the original height parameter is adjusted to obtain accurate height prediction parameters.

[0038] Furthermore, based on the meteorological influence weight, topographic influence weight, topographic compensation parameters, and meteorological compensation parameters, the original height parameters are adjusted to obtain accurate height prediction parameters, including:

[0039] Based on the meteorological influence weight, topographic influence weight, topographic compensation parameter, meteorological compensation parameter, smoothed original height parameter, and preset correction amplitude parameter, the accurate height prediction parameter is calculated using the following formula:

[0040]

[0041]

[0042] in, The time tag is Accurate height prediction parameters, The time tag is Smooth the original height parameter, It is a limiting function. The time tag is The influence weight of terrain The time tag is The sum of weights, The time tag is Terrain compensation parameters, The time tag is The weight of meteorological impact, The time tag is Meteorological compensation parameters, It is the minimum correction range. That is the maximum correction range. It is a non-zero coefficient.

[0043] Secondly, this application also provides a data prediction device for a radio altimeter, characterized in that the device comprises:

[0044] The data acquisition module is used to acquire the raw altitude parameters and motion data of the predicted target, as well as the real-time meteorological data of the environment in which the predicted target is located; the raw altitude parameters, motion data, and real-time meteorological data are all time-stamped; the motion data includes at least vertical velocity; the real-time meteorological data includes at least air pressure parameters and temperature parameters;

[0045] The height feature module is used to quantify and predict changes in the target's height based on the original height parameters and motion data, and obtain a height feature vector; the height feature vector includes at least a smoothed original height parameter with a time label;

[0046] The terrain determination module is used to determine the terrain type of the predicted target based on a preset terrain feature library and height feature vector, and obtain the terrain type label of the predicted target.

[0047] The altitude determination module is used to correct and smooth the original altitude parameters based on terrain type labels and real-time meteorological data to obtain accurate altitude prediction parameters.

[0048] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the data prediction method for any radio altimeter described in the first aspect of this application.

[0049] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the data prediction method for any radio altimeter described in the first aspect of this application.

[0050] The aforementioned data prediction method, apparatus, equipment, and medium for radio altimeters acquire the original altitude parameters and motion data of the target, as well as real-time meteorological data of the target's environment. All three data sets—original altitude parameters, motion data, and real-time meteorological data—are time-stamped. The motion data includes at least vertical velocity, and the real-time meteorological data includes at least air pressure and temperature parameters. Based on the original altitude parameters and motion data, the changes in the target's altitude are quantified to obtain an altitude feature vector. This feature vector includes at least a smoothed original altitude parameter with a time stamp. Based on a pre-defined terrain feature library and the altitude feature vector, the terrain type of the target is determined, resulting in a terrain type label. Based on the terrain type label and real-time meteorological data, the original smoothed altitude parameters are corrected to obtain accurate altitude prediction parameters. This effectively solves the measurement error problem caused by multipath effects in complex terrain using traditional radio altimeters, improving the reliability and accuracy of altitude data in low-altitude flight and complex terrain environments, and enhancing the reliability and accuracy of altitude measurement data for aircraft during low-altitude, complex terrain flight. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart illustrating a data prediction method for a radio altimeter provided in one embodiment of this application;

[0053] Figure 2 A flowchart illustrating a data prediction method for a radio altimeter provided in one embodiment of this application;

[0054] Figure 3 A schematic diagram of the structure of a data prediction device for a radio altimeter provided in one embodiment of this application;

[0055] Figure 4 This is a schematic diagram of the structure of a computer device for a data prediction method for a radio altimeter provided in one embodiment of this application. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0057] In one embodiment, such as Figure 1 As shown, a data prediction method for radio altimeters is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps S101-S104, wherein:

[0058] S101, acquire the raw altitude parameters and motion data of the target, as well as the real-time meteorological data of the environment in which the target is located; the raw altitude parameters, motion data and real-time meteorological data are all time-stamped; the motion data includes at least vertical velocity; the real-time meteorological data includes at least air pressure parameters and temperature parameters.

[0059] Specifically, the prediction target is the object whose altitude needs to be predicted. The prediction target can be any type of aircraft, such as an airplane. The raw altitude parameter is the altitude parameter directly output from the radio altimeter carried by the prediction target. The raw altitude parameter includes measurement bias caused by Gaussian white noise and multipath effects. Motion data is used to characterize the motion of the prediction target, including at least the vertical velocity in the vertical direction relative to the prediction target's coordinate system. Motion data can be obtained through the aircraft's inertial navigation system. Real-time meteorological data of the environment in which the prediction target is located includes at least air pressure and temperature parameters. The air pressure parameter can be obtained through the aircraft's static pressure sensor, and the temperature parameter can be measured and converted using a full-temperature sensor. The raw altitude parameter, motion data, and real-time meteorological data have been time-synchronized, and the time stamps on the raw altitude parameter, motion data, and real-time meteorological data will remain consistent. For example, the raw altitude parameter can be in the form of… ,in, Represents the time label The original altitude value at that time. Optionally, the motion data may also include a first directional acceleration, a second directional acceleration, a third directional acceleration, a pitch angle, and a roll angle, whereby the first directional acceleration characterizes the acceleration in the predicted target's forward direction. The second directional acceleration characterizes the acceleration in the predicted target's lateral direction. The third directional acceleration characterizes the acceleration in the predicted target's vertical direction.

[0060] S102, based on the original height parameters and motion data, quantify and predict the changes in the target's height to obtain a height feature vector; the height feature vector includes at least the smoothed original height parameters with time labels.

[0061] Specifically, the terminal filters the original height parameters based on motion data. This includes predicting and updating the original height parameters using predicted target dynamics data, effectively separating the actual height changes from measurement noise in the signal, outputting a smoothed original height parameter with a time label, and extracting the variation features of the smoothed original height parameter with the time label. These variation features are then combined with the smoothed original height parameter with the time label to form a height feature vector. The variation features mentioned above are divided into two categories: statistical features and frequency domain features. These features include, but are not limited to, the mean, variance, and spectral centroid of the smoothed original height parameter. The variation features also have time labels, representing data calculated based on the smoothed original height parameter with the corresponding time label. For example, the height feature vector can take the form of… .in, It is the termination sampling time label of the height feature vector. It is the initial sampling time label of the height feature vector. It's a time tag. and The mean of the smoothed original height parameter between them. It's a time tag. and The variance of the smoothed original height parameter between It's a time tag. and The spectral centroid of the smooth original height parameter between.

[0062] S103, based on the preset terrain feature library and height feature vector, determines the terrain type of the predicted target and obtains the terrain type label of the predicted target.

[0063] Specifically, the pre-set terrain feature library includes altitude feature vectors over different known terrains and their corresponding terrain labels, which can be collected based on actual work needs. The terminal uses a machine learning algorithm based on the pre-set terrain feature library to obtain a terrain classification model, so as to input the current altitude feature vector and obtain the current terrain type label. For example, the terrain type label includes a time label to indicate that the classification is based on the altitude feature vector with the corresponding time label. The terrain type label can take the form of… ,in, It is the termination sampling time label of the height feature vector. It is the initial sampling time label of the height feature vector. It is the index value of the terrain type label, where Used to characterize different terrain types, such as cities, plains, and mountains, time tags are placed in... and The terrain type label for any smoothed original height parameter between these parameters is... .

[0064] S104, based on terrain type labels and real-time meteorological data, corrects and smooths the original height parameters to obtain accurate height prediction parameters.

[0065] Specifically, the terminal obtains terrain compensation parameters by querying preset terrain height correction rules based on terrain type labels with time tags. It also obtains meteorological compensation parameters by correcting the smoothed original height parameters for each time tag using real-time meteorological data with time tags, based on the principle of atmospheric refraction. The terminal then weights and combines the terrain compensation parameters, meteorological compensation parameters, and smoothed original height parameters for the same time tag to obtain accurate height prediction parameters. The preset terrain height correction rules are based on an empirical value comparison table calibrated using electromagnetic simulation and historical flight data, storing corresponding error correction values ​​for different terrain categories, and can be incorporated into actual work processes.

[0066] This embodiment provides a data prediction method for radio altimeters. It establishes a unified time reference by synchronously acquiring time-stamped multi-source sensor data; it transforms the time-stamped raw altitude signal into a time-stamped terrain feature vector through filtering and feature extraction; it also utilizes a pre-set terrain feature library and machine learning model to achieve intelligent identification of the flight environment based on time synchronization, outputting time-stamped terrain types; and based on the time-stamped identification results and real-time meteorological data, it outputs accurate time-stamped altitude prediction parameters through adaptive weighted fusion. This effectively solves the measurement error problem caused by multipath effects in traditional radio altimeters under complex terrain, improving the reliability and accuracy of altitude data in low-altitude flight and complex terrain environments, and enhancing the reliability and accuracy of altitude measurement data for aircraft during low-altitude complex terrain flight.

[0067] In one embodiment, based on the original height parameters and motion data, the change in the target's height is quantified to obtain a height feature vector, including:

[0068] S201, based on motion data, filters the original height parameters to obtain smoothed original height parameters with time labels.

[0069] Specifically, the terminal uses time-stamped motion data, through the following formula: The absolute vertical acceleration corresponding to each time tag is calculated. The time tag is The absolute vertical acceleration. The time tag is The first directional acceleration is used to characterize the acceleration in the predicted direction of the target's movement. The time tag is The pitch angle. The time tag is The second directional acceleration is used to characterize the acceleration in the predicted lateral direction of the target. The time tag is Roll angle, The time tag is The third-order vertical acceleration is used to characterize the acceleration in the vertical direction of the predicted target. The terminal combines the time-stamped absolute vertical acceleration with a Kalman filter algorithm to filter the original height parameters, which also have time stamps, to obtain smoothed original height parameters corresponding to each time stamp. The Kalman filter algorithm is an optimal recursive state estimation algorithm. It predicts height changes by constructing a system state-space model and combining vertical acceleration information from motion data. The terminal uses a state vector containing height and vertical velocity to describe the system's dynamic characteristics through a state equation. It incorporates actual measurements into the estimation process using observation equations. Through iterative calculations in the prediction and update stages, it effectively separates the true height information from random noise in the signal, obtaining a smoothed height sequence. Each smoothed height value retains its original time stamp, ensuring time consistency in subsequent processing stages.

[0070] S202, based on the preset sampling time label parameters, combines each smooth original height parameter with a time label to obtain a sampling height parameter sequence.

[0071] Specifically, the terminal combines each smoothed original height parameter with a time tag based on a preset sampling time tag parameter to obtain a sampling height parameter sequence. The preset sampling time tag parameter defines the length of the sliding time window, used to extract a smoothed original height parameter containing a preset number of time tags from the continuous time series, thus obtaining the sampling height parameter sequence. Exemplarily, the preset sampling time tag parameter can be set according to the sampling frequency of the original height parameter; this embodiment does not further limit the setting of this preset sampling time tag parameter.

[0072] S203, based on the sampled height parameter sequence, calculate the mean of each smoothed original height parameter in the sampled height parameter sequence to obtain the smoothed height mean.

[0073] Specifically, the terminal calculates the arithmetic mean of each smoothed original height parameter in the sampled height parameter sequence to obtain the smoothed height mean. This mean reflects the average height level within the analysis time window and serves as a baseline for evaluating the amplitude of height fluctuations. The specific calculation process involves summing all smoothed height values ​​with time labels within the window and dividing by the window length to obtain the mean parameter, which also carries the latest time label of the window.

[0074] S204, based on the sampled height parameter sequence and smoothed height mean, quantifies the change characteristics of the target's height to obtain height change characteristic parameters; the height change characteristic parameters include height variance, height change slope, height difference mean, and height difference variance.

[0075] Specifically, the terminal calculates four height change characteristic parameters based on the sampled height parameter sequence and the smoothed height mean: height variance quantifies the overall fluctuation intensity by calculating the average of the squared differences between each smoothed original height parameter and the smoothed height mean; height change slope is obtained by performing linear regression analysis on the sampled height parameter sequence to obtain the trend line slope, reflecting the stable change trend of height; height difference mean characterizes the average rate of change by calculating the average of adjacent smoothed original height parameters; and height difference variance characterizes the instability of the rate of change by calculating the variance of the difference sequence.

[0076] S205, the sampled height parameter sequence is transformed in the frequency domain to obtain frequency domain height data, and the frequency domain features of the frequency domain height data are extracted to obtain frequency domain feature change parameters; the frequency domain feature change parameters include the spectral centroid and spectral entropy.

[0077] Specifically, the terminal performs frequency domain transformation on the sampled height parameter sequence to obtain frequency domain height data, and then extracts two frequency domain feature parameters: spectral centroid and spectral entropy. The Fast Fourier Transform (FFT) algorithm is a classic algorithm for converting time-domain signals into frequency-domain representations. It reveals the frequency structure characteristics of the signal by decomposing it into sinusoidal components of different frequencies. Specifically, the spectral centroid reflects the dominant frequency component by calculating the weighted average frequency of the spectrum, while the spectral entropy quantifies the spectral complexity by calculating the probability distribution entropy value of the spectral energy.

[0078] S206, based on the height feature change parameters, frequency domain feature change parameters, and smoothed original height parameters, obtains the height feature vector.

[0079] Specifically, the terminal integrates the height feature change parameters, the frequency domain feature change parameters, and the smoothed original height parameters with time labels to obtain a height feature vector.

[0080] This embodiment provides a data prediction method for radio altimeters. Through filtering and denoising, window segmentation, mean calculation, temporal feature extraction, frequency domain feature extraction, and feature integration, an information-rich altitude feature vector is ultimately formed. This method fully leverages the terrain information contained in the altitude data, transforming the original altitude sequence into a feature representation capable of effectively distinguishing different terrain patterns through multi-dimensional feature engineering. This provides rich and reliable input data for subsequent terrain identification, laying a solid foundation for the accuracy and reliability of the altitude correction process and enhancing the aircraft's adaptability to complex terrain environments.

[0081] In one embodiment, based on a preset terrain feature library and height feature vector, the terrain type of the predicted target is determined, and a terrain type label for the predicted target is obtained, including:

[0082] S301, based on the terrain feature library, inputs the height feature vector into the trained terrain classification model to obtain the terrain type and corresponding confidence of the predicted target, and obtains the terrain type label based on the terrain type and confidence.

[0083] Specifically, the terminal inputs the height feature vector into a trained terrain classification model to obtain the terrain type and corresponding confidence score of the predicted target. The terrain type and confidence score are then integrated to obtain the terrain type label. The terrain classification model is constructed using an SVM classifier algorithm based on a terrain feature library.

[0084] The trained terrain classification model was obtained through the following method:

[0085] S3011, Obtain the terrain feature library; the terrain feature library includes the height feature vectors of historical predicted targets and the corresponding terrain type labels.

[0086] Specifically, the terminal acquires a terrain feature database, which contains historical predicted target altitude feature vectors and corresponding terrain type labels. This data was collected through extensive preliminary flight tests. The construction of the terrain feature database requires systematic data collection flights over different known terrains. High-precision differential GPS or lidar measurements are used as ground truth references, while radio altimeter measurements and corresponding actual terrain types are recorded simultaneously. Then, a standardized altitude feature vector is obtained through a feature extraction process, ultimately forming a set of sample pairs: {altitude feature vector, terrain type label}.

[0087] S3012 divides the terrain feature library into a model training set and a model validation set according to a preset ratio.

[0088] Specifically, the terminal divides the terrain feature library into a model training set and a model validation set according to a preset ratio. The training set is used for learning and optimizing model parameters, while the validation set is used to evaluate the model's generalization performance and prevent overfitting. For example, the preset division ratio usually follows standard practices in the field of machine learning, and is set by default to a model training set: model validation set ratio of 8:2. This embodiment does not further limit the setting of this preset division ratio.

[0089] S3013, based on the model training set and a preset kernel function, constructs an initial terrain classification model with height feature vector as input and terrain category label as prediction target.

[0090] Specifically, the terminal constructs an initial terrain classification model based on the model training set and a preset kernel function, using height feature vectors as input and terrain category labels as the prediction target. Specifically, the terminal uses an optimization algorithm based on the preset kernel function to learn support vectors and their corresponding coefficients from the model training set, maximizing the margin between samples of different categories in the feature space, thus obtaining the initial terrain classification model. The Support Vector Machine (SVM) algorithm finds the maximum margin classification hyperplane by solving a convex optimization problem. For example, the preset kernel function is typically a radial basis function; this embodiment does not further limit the setting of this preset kernel function.

[0091] S3014, the accuracy of the initial terrain classification model is verified by the model validation set. When the accuracy of the initial terrain classification model reaches the preset threshold, the terrain classification model is obtained.

[0092] Specifically, the terminal verifies the accuracy of the initial terrain classification model using a model validation set. When the accuracy of the initial terrain classification model reaches a preset threshold, the final terrain classification model is obtained. The accuracy verification process involves inputting the height feature vectors from the validation set into the trained initial model, comparing the terrain type predicted by the model with the actual terrain label, and calculating the overall classification accuracy. For example, the preset threshold is set to 90% by default, but it can be set according to actual work needs. This embodiment does not further limit the setting of this preset threshold.

[0093] This embodiment provides a data prediction method for radio altimeters. It constructs a terrain feature library by collecting historical flight data, divides the data into training and validation sets, builds an initial classification model based on the training set, evaluates the model performance using the validation set to determine the terrain classification model, and then uses the aforementioned terrain classification model to classify real-time altitude feature vectors, outputting terrain type labels with confidence levels. This ensures the accuracy and reliability of terrain classification, establishes an accurate mapping relationship between complex terrain features and altitude change patterns, enables automatic identification of the flight environment and intelligent altitude correction accordingly, effectively solves the systematic error problem caused by multipath effects in complex terrain using traditional radio altimeters, improves the reliability and accuracy of altitude measurement in low-altitude flight and complex terrain environments, and provides important technical support for flight safety.

[0094] In one embodiment, based on terrain type labels and real-time weather data, the original height parameters are corrected and smoothed to obtain accurate height prediction parameters, including:

[0095] S401, based on the preset terrain height correction rules and terrain type labels, obtains terrain compensation parameters; the terrain compensation parameters are labeled with time.

[0096] Specifically, the terminal obtains the terrain compensation parameters corresponding to each time tag based on preset terrain height correction rules and terrain type tags with time stamps using a lookup table method. The preset terrain height correction rules are a pre-calibrated lookup table storing systematic error compensation values ​​corresponding to different terrain types. These compensation values ​​are obtained through analysis of a large amount of historical flight data and electromagnetic simulation calculations, reflecting the characteristics of radio altimeter measurement deviations caused by factors such as multipath effects under specific terrain conditions. The preset terrain height correction rules can be set according to actual work needs; this embodiment does not further limit the setting of these preset rules. For example, when the terrain type tag is identified as a flat plain, the terrain compensation parameters are usually close to zero because flat terrain minimizes the measurement error caused by radio wave reflection; when the terrain type tag is identified as a mountainous area, the terrain compensation parameters may be negative to correct for errors caused by overestimating the measurement value due to reflection from the slope; when the terrain type tag is identified as an urban built-up area, the terrain compensation parameters may be positive to compensate for errors caused by underestimating the measurement value due to near-distance reflection from building tops. Among them, the terrain compensation parameters are labeled with time tags to indicate that they are obtained by querying the terrain type label according to the corresponding time tag. The terrain compensation parameters of the smoothed original height parameters contained in the same height feature vector are generally the same in value.

[0097] S402, based on real-time meteorological data, quantifies the impact of real-time meteorological data on smoothing the original height parameter to obtain meteorological compensation parameters; the meteorological compensation parameters are labeled with time.

[0098] Specifically, the terminal calculates meteorological compensation parameters corresponding to each time tag based on real-time meteorological data with time tags, according to the principle of atmospheric refraction. The principle of atmospheric refraction is a physical principle describing the curvature of radio wave propagation paths in the atmosphere. It quantifies the combined impact of changes in radio wave propagation speed and path curvature on altitude measurement by calculating the atmospheric refractive index and its vertical gradient. The meteorological compensation parameters are time-tagged to indicate that they are calculated based on real-time meteorological data for the corresponding time tag.

[0099] S403 adjusts the original height parameters based on terrain type labels, real-time meteorological data, terrain compensation parameters, and meteorological compensation parameters to obtain accurate height prediction parameters.

[0100] Specifically, the terminal calculates the weights of terrain compensation parameters and meteorological compensation parameters based on terrain type labels and real-time meteorological data, and applies the terrain compensation parameters and meteorological compensation parameters to the smoothed original height parameters of the corresponding time label according to the weights, so as to obtain the accurate height prediction parameters of the corresponding time label.

[0101] This embodiment provides a data prediction method for radio altimeters. It obtains terrain compensation parameters by querying preset compensation rules based on terrain identification results, and calculates meteorological compensation parameters based on real-time meteorological data using atmospheric refraction principles. Furthermore, it integrates these two types of compensation parameters into a smoothed altitude parameter using an adaptive weighted fusion algorithm. This method fully considers the impact of complex terrain and variable meteorological conditions on radio altimeter measurements. By combining multi-source data fusion and physical model compensation, it effectively solves the systematic measurement error problem caused by multipath effects in complex terrains such as mountainous and urban areas, while maintaining altitude measurement accuracy under various atmospheric conditions. This improves the reliability and accuracy of altitude measurements in low-altitude flight and complex terrain environments, providing important technical support for flight safety.

[0102] In one embodiment, based on real-time meteorological data, the impact of real-time meteorological data on smoothing the original height parameter is quantified to obtain meteorological compensation parameters, including:

[0103] S501 calculates the refractive index parameters of the environment in which the target is located based on air pressure and temperature parameters.

[0104] Specifically, for each time tag corresponding to the air pressure and temperature parameters, the terminal uses the following formula: The refractive index parameter corresponding to this time stamp is calculated. The time tag is The refractive index parameter is used to characterize the ability of the current atmospheric environment to refract radio waves. The time tag is Temperature parameters. The time tag is The air pressure parameters. These are preset water vapor pressure parameters, which can be set according to actual working conditions. This embodiment does not further limit the setting of these preset water vapor pressure parameters.

[0105] S502, the gradient of the refractive index parameter is calculated to obtain the vertical gradient of the refractive index; and based on the vertical gradient of the refractive index, the equivalent Earth radius factor is calculated.

[0106] Specifically, the terminal performs gradient calculations on the refractive index parameters corresponding to each time tag to obtain the vertical gradient of the refractive index, and uses the following formula based on the vertical gradient of the refractive index: The equivalent Earth radius factor was calculated. The time tag is The equivalent Earth radius factor is used to characterize the ratio between the virtual Earth radius (used to simplify the calculation of radio wave propagation paths) and the actual Earth radius. The time tag is The refractive index parameter is used to calculate the gradient of the vertical height to obtain the refractive index vertical gradient corresponding to each time tag. It is the actual radius of the Earth.

[0107] S503 calculates meteorological compensation parameters based on the equivalent Earth radius factor, the preset atmospheric proportion factor, and the smoothed original height parameters.

[0108] Specifically, for each time tag, the terminal uses the following formula for the original altitude parameter, equivalent Earth radius factor, and preset atmospheric scale factor: The meteorological compensation parameters for the corresponding time label are calculated. The time tag is Meteorological compensation parameters. The time tag is The smoothed original height parameter. The time tag is The equivalent Earth radius factor. It is a preset atmospheric scaling factor used to compensate for errors caused by model simplification and to adapt to regional atmospheric characteristics in different regions. Its value is usually determined through regression analysis of a large amount of flight test data, and the value range is generally between 0.8 and 1.2. It can be set according to actual work. This embodiment does not further limit the setting of this preset atmospheric scaling factor.

[0109] This embodiment provides a data prediction method for radio altimeters. By calculating atmospheric refractive index parameters based on real-time air pressure and temperature data, it accurately characterizes the current atmospheric radio wave propagation characteristics. Furthermore, through vertical gradient analysis and equivalent Earth radius calculation, it quantifies the bending effect of atmospheric refraction on the radio wave propagation path. It also combines a preset atmospheric scaling factor and the original altitude measurement value to calculate precise meteorological compensation parameters. This method fully considers the physical characteristics of radio wave propagation under different atmospheric conditions, effectively compensating for altitude measurement errors caused by atmospheric refraction. It improves the reliability and accuracy of altitude measurement in low-altitude flight and complex terrain environments, providing crucial technical support for flight safety.

[0110] In one embodiment, based on terrain type labels, real-time meteorological data, terrain compensation parameters, and meteorological compensation parameters, the original height parameters are adjusted to obtain accurate height prediction parameters, including:

[0111] S601, based on the terrain type label, normalizes the confidence score to obtain the terrain influence weight.

[0112] Specifically, the terminal normalizes the confidence scores contained in the terrain type labels to obtain the terrain influence weights. The sigmoid function can be used to normalize the confidence scores, resulting in terrain influence weights with time labels. For the same height feature vector containing smoothed original height parameters, the terrain influence weights are numerically identical.

[0113] S602, based on real-time meteorological data, quantifies the fluctuation of real-time meteorological data to obtain the standard deviation of meteorological data; the standard deviation of meteorological data includes the standard deviation of air pressure and the standard deviation of temperature.

[0114] Specifically, the terminal statistically analyzes and quantifies the fluctuations in meteorological data for air pressure and temperature parameters at different time tags, obtaining the meteorological data standard deviation, which includes the standard deviations of air pressure and temperature. The air pressure standard deviation reflects the degree of fluctuation in recent air pressure measurements, while the temperature standard deviation reflects the degree of fluctuation in recent temperature measurements.

[0115] S603 calculates the meteorological impact weights based on the standard deviation of meteorological data and real-time meteorological data.

[0116] Specifically, based on preset standard air pressure and temperature parameters, as well as the standard deviations of air pressure and temperature, the terminal uses the following formula to calculate the real-time meteorological data for each time tag: The meteorological influence weights of the smoothed original altitude parameters for the corresponding time labels are calculated. The time tag is The weight of meteorological impact, The time tag is air pressure parameters, These are standard atmospheric pressure parameters. It is the standard deviation of air pressure. The time tag is Temperature parameters, These are standard temperature parameters. This refers to the temperature standard deviation. For example, the preset standard pressure and standard temperature parameters can be set according to the pressure and temperature values ​​of standard atmospheric conditions in actual work.

[0117] S604 adjusts the original height parameters based on meteorological influence weight, topographic influence weight, topographic compensation parameters, and meteorological compensation parameters to obtain accurate height prediction parameters.

[0118] Specifically, the terminal performs a weighted combination of the smoothed raw altitude parameters based on the air pressure and temperature values ​​of the standard atmospheric conditions for each time tag, to obtain accurate altitude prediction parameters for each time tag.

[0119] This embodiment provides a data prediction method for radio altimeters. It calculates the terrain influence weight based on terrain classification confidence levels to quantify the reliability of terrain identification results; it obtains the meteorological data standard deviation through statistical analysis of meteorological data fluctuations to assess the stability of meteorological conditions; it calculates the meteorological influence weight by combining the meteorological data standard deviation and real-time meteorological conditions to quantify the reliability of meteorological compensation; and it performs amplitude-limited weighted fusion of terrain compensation parameters and meteorological compensation parameters based on the two weights to obtain accurate altitude prediction parameters. Through a dynamic weight allocation mechanism, it achieves adaptive optimization fusion of multi-source compensation quantities under different environmental conditions, effectively solving the problem of poor adaptability of traditional fixed-weight fusion methods in complex environments, and improving the measurement accuracy and reliability of radio altimeters under varying terrain and meteorological conditions.

[0120] In one embodiment, the original height parameters are adjusted based on meteorological influence weights, terrain influence weights, terrain compensation parameters, and meteorological compensation parameters to obtain accurate height prediction parameters, including:

[0121] Based on the meteorological influence weight, topographic influence weight, topographic compensation parameter, meteorological compensation parameter, smoothed original height parameter, and preset correction amplitude parameter, the accurate height prediction parameter is calculated using the following formula:

[0122]

[0123]

[0124] in, The time tag is Accurate height prediction parameters, The time tag is Smooth the original height parameter, It is a limiting function. The time tag is The influence weight of terrain The time tag is The sum of weights, The time tag is Terrain compensation parameters, The time tag is The weight of meteorological impact, The time tag is Meteorological compensation parameters, It is the minimum correction range. That is the maximum correction range. It is a non-zero coefficient.

[0125] Specifically, the terminal adjusts the smoothed original altitude parameters for the corresponding time label based on preset correction parameters, including meteorological influence weight, terrain influence weight, terrain compensation parameters, and meteorological compensation parameters, to obtain accurate altitude prediction parameters for that time label. The time label is... Smoothed original height parameters It can be obtained from the height feature vector. (Limiting function) Used to ensure that accurate altitude prediction parameters are within safe limits. Time stamps are... Terrain influence weight This can be obtained from S601. The time stamp is... The total weight The time tag is Terrain influence weight Meteorological impact weight Non-zero coefficients The sum is obtained by addition. The time stamp is... Meteorological impact weight This can be obtained from S603. The time stamp is... Terrain compensation parameters This can be obtained from S401. The time stamp is... Meteorological compensation parameters This can be obtained from S503. Minimum correction magnitude. and maximum correction range This parameter is used to limit the variation range of accurate height prediction parameters and smooth the original height parameters. It can be set according to actual work requirements; this embodiment does not further limit the setting of this preset correction range parameter. (Non-zero coefficient) To prevent the denominator from being zero, the setting can be configured according to the actual operation. For example, a limiting function... The form can be .

[0126] This embodiment provides a data prediction method for radio altimeters. By establishing a complete weighted fusion and amplitude limiting protection mechanism, it optimally fuses terrain compensation parameters and meteorological compensation parameters according to their respective influence weights, and precisely adjusts the smoothed original altitude parameters. The weight calculation fully considers the confidence level of terrain classification and the reliability of meteorological data, and the amplitude limiting function ensures the safety of the correction process. The obtained accurate altitude prediction parameters effectively overcome the systematic measurement errors of traditional radio altimeters in complex terrain and variable weather environments, and improve the measurement accuracy and reliability of radio altimeters under variable terrain and weather conditions.

[0127] To further illustrate the scheme for obtaining terrain type labels for predicted targets, terrain can also be identified from images to generate terrain type labels, as described above. Figure 2This includes the following steps:

[0128] S01, Image data acquisition and preprocessing.

[0129] Specifically, this includes: the terminal acquiring observational image data of the terrain where the predicted target is located. This observational image data is obtained by directly capturing the surface scenery below the aircraft using an image sensor of the predicted target. For example, the image sensor can be an electro-optical camera, such as a visible light or infrared band sensor. The image sensor simultaneously acquires a sequence of surface images to obtain observational image data. Each frame of the observational image data... All data sampling periods are strictly aligned with those of the radio altimeter and contain the same time stamps. To ensure temporal consistency of multi-source data, the terminal also performs geometric distortion correction, illumination normalization, and format standardization on the observed image data to obtain preprocessed sampled image data.

[0130] S02, Feature Extraction and Terrain Classification.

[0131] Specifically, the terminal uses a pre-trained convolutional neural network (CNN) model to process the aforementioned image tensors. A CNN is a deep learning model specifically designed for processing grid-like data (such as images). Through multiple convolutional layers, pooling layers, and non-linear activation functions, it can automatically extract hierarchical features from raw pixels, progressively extracting features ranging from edges and textures to local patterns and even semantic objects. In this application, the trained CNN model uses the aforementioned pre-processed image... As input, after forward propagation computation, a probability distribution vector is finally output through a Softmax classification layer. Each component of the vector Each probability corresponds to a preset terrain type (such as "flat plain", "undulating hills", "urban built-up area", "water area", "forest", etc.), and the magnitude of the probability value represents the confidence level of the model in believing that the current image belongs to that category.

[0132] S03, Terrain Label Generation.

[0133] Specifically, the process includes: the terminal selecting the category with the highest predicted probability from the probability distribution vector to obtain the terrain type; and selecting the predicted probability corresponding to the terrain type as the corresponding confidence level. The terminal then concatenates the terrain type and the corresponding confidence level to obtain the terrain type label of the predicted target.

[0134] The aforementioned data prediction method for radio altimeters involves acquiring the original altitude parameters and motion data of the target, as well as real-time meteorological data of the target's environment. All three data sets include time stamps. The motion data includes at least vertical velocity, and the real-time meteorological data includes at least air pressure and temperature parameters. Based on the original altitude parameters and motion data, the changes in the target's altitude are quantified to obtain an altitude feature vector. This vector includes at least a smoothed original altitude parameter with a time stamp. Based on a pre-defined terrain feature library and the altitude feature vector, the terrain type of the target is determined, resulting in a terrain type label. Based on the terrain type label and real-time meteorological data, the original altitude parameters are corrected and smoothed to obtain accurate altitude prediction parameters. This method effectively solves the measurement error problem caused by multipath effects in complex terrain using traditional radio altimeters, improving the reliability and accuracy of altitude data in low-altitude flight and complex terrain environments, and enhancing the reliability and accuracy of altitude measurement data for aircraft during low-altitude flight in complex terrain.

[0135] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0136] Based on the same inventive concept, this application also provides a data prediction device for a radio altimeter to implement the data prediction method for radio altimeters described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the data prediction device for radio altimeters provided below can be found in the limitations of the data prediction method for radio altimeters described above, and will not be repeated here.

[0137] In one exemplary embodiment, such as Figure 3 As shown, a data prediction device 200 for a radio altimeter is provided, comprising:

[0138] The data acquisition module 201 is used to acquire the original height parameters and motion data of the predicted target, as well as the real-time meteorological data of the environment in which the predicted target is located; the original height parameters, motion data and real-time meteorological data are all time-stamped; the motion data includes at least vertical velocity; the real-time meteorological data includes at least air pressure parameters and temperature parameters;

[0139] The height feature module 202 is used to quantify and predict the change in the height of the target based on the original height parameters and motion data, and obtain a height feature vector; the height feature vector includes at least the smoothed original height parameters with time labels;

[0140] The terrain determination module 203 is used to determine the terrain type of the predicted target based on a preset terrain feature library and height feature vector, and to obtain the terrain type label of the predicted target.

[0141] The height determination module 204 is used to correct and smooth the original height parameters based on terrain type labels and real-time meteorological data to obtain accurate height prediction parameters.

[0142] Furthermore, the height feature module includes:

[0143] The smoothing unit is used to filter the original height parameters based on motion data to obtain smoothed original height parameters with time labels.

[0144] The time segmentation unit is used to combine each smoothed original height parameter with a time label based on a preset sampling time label parameter to obtain a sampling height parameter sequence;

[0145] The mean calculation unit is used to calculate the mean of each smoothed original height parameter in the sampled height parameter sequence based on the sampled height parameter sequence, so as to obtain the smoothed height mean.

[0146] The temporal feature calculation unit is used to quantify the change characteristics of the target's altitude based on the sampled altitude parameter sequence and the smoothed altitude mean, and obtain altitude change feature parameters; the altitude feature change parameters include altitude variance, altitude change slope, altitude difference mean, and altitude difference variance;

[0147] The frequency domain feature calculation unit is used to perform frequency domain transformation on the sampled height parameter sequence to obtain frequency domain height data, and extract the frequency domain features of the frequency domain height data to obtain frequency domain feature change parameters; the frequency domain feature change parameters include spectral centroid and spectral entropy;

[0148] The feature vector component is used to obtain the height feature vector based on the height feature variation parameters, frequency domain feature variation parameters, and smoothed original height parameters.

[0149] Furthermore, the smoothing unit is also used for:

[0150] For each time tag corresponding to the motion data, the absolute vertical acceleration corresponding to each time tag is calculated based on the first direction acceleration, the second direction acceleration, the third direction acceleration, the pitch angle and the roll angle.

[0151] For the motion data and original height parameters corresponding to the same time label, initial state parameters for each time label are constructed based on the vertical velocity and original height parameters; the initial state parameters include the initial state vector and the initial covariance matrix.

[0152] Calculate the difference between each time tag and construct the state transition matrix for each time tag based on the difference;

[0153] For each time tag, the initial state parameters of the previous time tag are updated based on the state transition matrix and absolute vertical acceleration of the time tag to obtain the predicted state parameters of each time tag; the predicted state parameters include the predicted state vector and the predicted covariance matrix.

[0154] Based on the predicted covariance matrix of each time tag and the preset observation matrix, the gain value of each time tag is calculated.

[0155] For each time tag, the predicted state vector is corrected based on the gain value of the time tag and the original height parameter to obtain the smoothed original height parameter with time tag.

[0156] Furthermore, the terrain determination module is also used for:

[0157] Based on the terrain feature library, the height feature vector is input into the trained terrain classification model to obtain the terrain type and corresponding confidence of the predicted target, and the terrain type label is obtained based on the terrain type and confidence.

[0158] The trained terrain classification model was obtained through the following method:

[0159] Obtain the terrain feature library; the terrain feature library includes the height feature vectors of historical predicted targets and the corresponding terrain type labels;

[0160] The terrain feature database is divided into a model training set and a model validation set according to a preset ratio;

[0161] Based on the model training set and the preset kernel function, an initial terrain classification model is constructed with height feature vector as input and terrain category label as prediction target;

[0162] The accuracy of the initial terrain classification model is verified using a model validation set. When the accuracy of the initial terrain classification model reaches a preset threshold, the terrain classification model is obtained.

[0163] Furthermore, the altitude determination module includes:

[0164] The terrain compensation calculation unit is used to obtain terrain compensation parameters based on preset terrain height correction rules and terrain type labels; the terrain compensation parameters are labeled with time.

[0165] The meteorological compensation calculation unit is used to quantify the impact of real-time meteorological data on the smoothing of the original height parameter based on real-time meteorological data, and obtain meteorological compensation parameters; the meteorological compensation parameters are labeled with time.

[0166] The accurate height calculation unit is used to adjust the smoothed original height parameters based on terrain type labels, real-time meteorological data, terrain compensation parameters, and meteorological compensation parameters to obtain accurate height prediction parameters.

[0167] Furthermore, the meteorological compensation calculation unit is also used for:

[0168] Based on the air pressure and temperature parameters, the refractive index parameters of the environment in which the target is located are calculated.

[0169] The refractive index parameter is gradient calculated to obtain the vertical refractive index gradient; and based on the vertical refractive index gradient, the equivalent Earth radius factor is calculated.

[0170] Meteorological compensation parameters are calculated based on the equivalent Earth radius factor, the preset atmospheric proportion factor, and the smoothed original height parameters.

[0171] Furthermore, the accurate height calculation unit also includes:

[0172] The terrain weight calculation subunit is used to normalize the confidence score based on the terrain type label to obtain the terrain influence weight.

[0173] The temperature adjustment calculation subunit is used to quantify the fluctuation of real-time meteorological data based on real-time meteorological data and obtain the standard deviation of meteorological data; the standard deviation of meteorological data includes the standard deviation of air pressure and the standard deviation of temperature.

[0174] The temperature weight calculation subunit is used to calculate the meteorological impact weight based on the standard deviation of meteorological data and real-time meteorological data.

[0175] The original parameter correction subunit is used to adjust the smoothed original height parameters based on meteorological influence weight, terrain influence weight, terrain compensation parameters, and meteorological compensation parameters to obtain accurate height prediction parameters.

[0176] Furthermore, the original parameter correction subunit is also used for:

[0177] Based on the meteorological influence weight, topographic influence weight, topographic compensation parameter, meteorological compensation parameter, smoothed original height parameter, and preset correction amplitude parameter, the accurate height prediction parameter is calculated using the following formula:

[0178]

[0179]

[0180] in, The time tag is Accurate height prediction parameters, The time tag is Smooth the original height parameter, It is a limiting function. The time tag is The influence weight of terrain The time tag is The sum of weights, The time tag is Terrain compensation parameters, The time tag is The weight of meteorological impact, The time tag is Meteorological compensation parameters, It is the minimum correction range. That is the maximum correction range. It is a non-zero coefficient.

[0181] In one embodiment, such as Figure 3 A computer device is provided, comprising:

[0182] At least one processor 301, and a memory 302 communicatively connected to at least one of the processors 301: the memory stores application code executable by at least one of the processors, the application code being executed by at least one of the processors to enable at least one of the processors to perform the data prediction method for the radio altimeter as described above.

[0183] Computer equipment may also include: sensor 303.

[0184] The processor 301, memory 302 and sensor 303 can be connected via a bus or other means, with the bus being an example in the figure.

[0185] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0186] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0187] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for predicting data from a radio altimeter, characterized in that, The method includes: The system acquires the raw altitude parameters and motion data of the target, as well as real-time meteorological data of the environment in which the target is located; the raw altitude parameters, the motion data, and the real-time meteorological data are all time-stamped; the motion data includes at least vertical velocity; and the real-time meteorological data includes at least air pressure and temperature parameters. Based on the original height parameters and the motion data, the change in the height of the predicted target is quantified to obtain a height feature vector; the height feature vector includes at least the smoothed original height parameters with the time label. Based on the preset terrain feature library and the height feature vector, the terrain type of the predicted target is determined, and the terrain type label of the predicted target is obtained. Based on the terrain type label and the real-time meteorological data, the smoothed original height parameter is corrected to obtain accurate height prediction parameters.

2. The method according to claim 1, characterized in that, The step of quantifying the change in the predicted target's altitude based on the original altitude parameters and the motion data to obtain an altitude feature vector includes: Based on the motion data, the original height parameters are filtered to obtain smoothed original height parameters with the time labels. Based on the preset sampling time tag parameters, the smoothed original height parameters with the time tags are combined to obtain a sampling height parameter sequence; Based on the sampled height parameter sequence, the mean of each smoothed original height parameter in the sampled height parameter sequence is calculated to obtain the smoothed height mean. Based on the sampled height parameter sequence and the smoothed height mean, the variation characteristics of the predicted target's height are quantified to obtain height variation feature parameters; the height variation feature parameters include height variance, height variation slope, height difference mean, and height difference variance; The sampled height parameter sequence is transformed in the frequency domain to obtain frequency domain height data, and the frequency domain features of the frequency domain height data are extracted to obtain frequency domain feature change parameters; the frequency domain feature change parameters include spectral centroid and spectral entropy; The height feature vector is obtained based on the height feature change parameters, the frequency domain feature change parameters, and the smoothed original height parameters.

3. The method according to claim 1, characterized in that, The method of determining the terrain type of the predicted target based on a preset terrain feature library and the height feature vector, and obtaining the terrain type label of the predicted target, includes: Based on the terrain feature library, the height feature vector is input into the trained terrain classification model to obtain the terrain type and corresponding confidence level of the predicted target, and the terrain type label is obtained based on the terrain type and the confidence level. The trained terrain classification model was obtained through the following method: Obtain the terrain feature library; the terrain feature library includes the height feature vector of the historical predicted target and the corresponding terrain type label; The terrain feature library is divided into a model training set and a model validation set according to a preset ratio; Based on the model training set and the preset kernel function, an initial terrain classification model is constructed using the height feature vector as input and the terrain category label as the prediction target. The accuracy of the initial terrain classification model is verified using the model validation set. When the accuracy of the initial terrain classification model reaches a preset threshold, the terrain classification model is obtained.

4. The method according to claim 3, characterized in that, The process of correcting the smoothed original height parameters based on the terrain type label and the real-time meteorological data to obtain accurate height prediction parameters includes: Based on the preset terrain height correction rules and the terrain type label, terrain compensation parameters are obtained; the terrain compensation parameters are labeled with the time label. Based on the real-time meteorological data, the impact of the real-time meteorological data on the smoothed original height parameter is quantified to obtain meteorological compensation parameters; the meteorological compensation parameters are labeled with the time tag. Based on the terrain type label, the real-time meteorological data, the terrain compensation parameters, and the meteorological compensation parameters, the smoothed original height parameters are adjusted to obtain the accurate height prediction parameters.

5. The method according to claim 4, characterized in that, The step of quantifying the impact of the real-time meteorological data on the smoothed original height parameter based on the real-time meteorological data to obtain meteorological compensation parameters includes: Based on the air pressure parameter and the temperature parameter, the refractive index parameter of the environment in which the predicted target is located is calculated; The refractive index parameter is subjected to gradient calculation to obtain the vertical gradient of the refractive index; and the equivalent Earth radius factor is calculated based on the vertical gradient of the refractive index. The meteorological compensation parameters are calculated based on the equivalent Earth radius factor, the preset atmospheric proportion factor, and the smoothed original height parameters.

6. The method according to claim 4, characterized in that, The process of adjusting the smoothed original height parameters based on the terrain type label, the real-time meteorological data, the terrain compensation parameters, and the meteorological compensation parameters to obtain the accurate height prediction parameters includes: Based on the terrain type label, the confidence score is normalized to obtain the terrain influence weight. Based on the real-time meteorological data, the fluctuation of the real-time meteorological data is quantified to obtain the standard deviation of the meteorological data; the standard deviation of the meteorological data includes the standard deviation of air pressure and the standard deviation of temperature. Based on the standard deviation of the meteorological data and the real-time meteorological data, the meteorological impact weight is calculated. Based on the meteorological influence weight, the terrain influence weight, the terrain compensation parameter, and the meteorological compensation parameter, the smoothed original height parameter is adjusted to obtain the accurate height prediction parameter.

7. The method according to claim 6, characterized in that, The process of adjusting the smoothed original height parameters based on the meteorological influence weight, the terrain influence weight, the terrain compensation parameter, and the meteorological compensation parameter to obtain the accurate height prediction parameters includes: Based on the meteorological influence weight, the terrain influence weight, the terrain compensation parameter, the meteorological compensation parameter, the smoothed original height parameter, and the preset correction amplitude parameter, the accurate height prediction parameter is calculated using the following formula: in, The time tag is Accurate height prediction parameters, The time tag is Smooth the original height parameter, It is a limiting function. The time tag is The influence weight of terrain The time tag is The sum of weights, The time tag is Terrain compensation parameters, The time tag is The weight of meteorological impact, The time tag is Meteorological compensation parameters, It is the minimum correction range. That is the maximum correction range. It is a non-zero coefficient.

8. A data prediction device for a radio altimeter, characterized in that, The device includes: The data acquisition module is used to acquire the original altitude parameters and motion data of the predicted target, as well as the real-time meteorological data of the environment in which the predicted target is located; the original altitude parameters, the motion data, and the real-time meteorological data are all time-stamped; the motion data includes at least vertical velocity; the real-time meteorological data includes at least air pressure parameters and temperature parameters; The height feature module is used to quantify the change in the height of the predicted target based on the original height parameters and the motion data, and obtain a height feature vector; the height feature vector includes at least a smoothed original height parameter with the time label; The terrain determination module is used to determine the terrain type of the predicted target based on a preset terrain feature library and the height feature vector, and to obtain the terrain type label of the predicted target. The altitude determination module is used to correct the smoothed original altitude parameters based on the terrain type label and the real-time meteorological data to obtain accurate altitude prediction parameters.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images